Privacy-Preserving Prediction of Electricity Consumption in Smart Grids
4th International Conference on Smart Grid and Renewable Energy, SGRE 2024, Doha, Qatar, 8 - 10 Ocak 2024, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/sgre59715.2024.10428900
- Basıldığı Şehir: Doha
- Basıldığı Ülke: Qatar
- Anahtar Kelimeler: differential privacy, electricity consumption prediction, GRU, LSTM, smart grid
- Ankara Yıldırım Beyazıt Üniversitesi Adresli: Hayır
Özet
Smart grid is a large-scale electrical network that allows the network administrator to efficiently manage the entire system using data-driven methods. Smart meters serve as a crucial data source in a smart grid system. They accurately measure the electricity use of households and transfer this data to the control center. Periodically, specialists utilize smart meter data to predict electricity demand. This prediction empowers smart grid operators to assign the necessary quantity of electricity and effectively operate the system. Although, as electricity usage is considered as personal data, it is imperative to consider implementing protective measures. This research presents a novel method for predicting electricity consumption while taking data privacy into account. We ensured data privacy by implementing differential privacy. Besides, prediction based experiments are conducted with two mostly used RNN based deep learning methods, LSTM and GRU. Experimental results have shown that the model proposed in this study produces effective results for utility, privacy and prediction.